Toshiba Satellite M505 overheating problem

[almikul]

According to https://bugs.launchpad.net/bugs/500666, the fan would start when suspending, then resuming.


Can you confirm this?

Yes, it is true for M505.

[wafflemelon]

My fan is finally working, with the following boot option: acpi.power_nocheck=1

Unfortunately, it only turns on, at a low speed, when approaching 70C.

So still no problem solved.

Of course, acpi=off does wonders for the fan, but in turn, I have to run X at lo-res and have no wifi.

[wafflemelon]

OK, I got the fans to work rather reasonably. I will see how it goes, but for now, here's what I did on 9.10 :

* boot with option
acpi.power_nocheck=1

* Install & configure powersaved: http://ubuntuforums.org/showthread.php?t=846480

* Toshiba U500 users found that having the graphics drivers installed has helped. Try that also.

In the end, all that I managed is to have the trip points upheld. The ideal is setting our own trip points, but I have no idea how we can do this.

Currently posting this from Karmic, at 53 C

[wafflemelon]

I fiddled around with DSDT today, and I might have got it right. Unfortunately, it seems the kernel devs once more burst our bubble: Custom DSDT support has recently been disabled.

The only solution is to recompile the kernel with the DSDT.
I spent the last hour and a half compiling the kernel on this laptop which already has thermal problems, to have the whole process fail.

If anyone is willing to try it and perhaps make a patch for all of us, I'll gladly submit the custom DSDT.

Funny thing is, Windows allows me to fiddle with my trip points AND use a custom DSDT (two solutions for our thermal issues). Since when does Linux treat it's users like children?

[almikul]

OK, I got the fans to work rather reasonably. I will see how it goes, but for now, here's what I did on 9.10 :

* boot with option
acpi.power_nocheck=1

* Install & configure powersaved: http://ubuntuforums.org/showthread.php?t=846480

* Toshiba U500 users found that having the graphics drivers installed has helped. Try that also.

In the end, all that I managed is to have the trip points upheld. The ideal is setting our own trip points, but I have no idea how we can do this.

Did not work for me. Here is the output of thermal trip points:

critical (S5):           108 C
passive:                 101 C: tc1=30 tc2=30 tsp=50 devices=CPU0 CPU1 
active[0]:               98 C: devices=FAN0 
active[1]:               82 C: devices=FAN1 
active[2]:               72 C: devices=FAN2 
active[3]:               62 C: devices=FAN3 
active[4]:               57 C: devices=FAN4 
active[5]:               37 C: devices=FAN5

I tried to put custom trip points using powersaved, but it does not seem that anything is affected at all. I still have to suspend and resume to make my fan working.

[wafflemelon]

Yes, powersave doesn't override the trip points. But somehow after configuring it, the fan started to work.

Also, your trip points are different from mine. I also get different trip points for different kernel acpi options.

The DSDT on our laptops has different configurations according to the OS it detects. It discriminates the following strings:

"Linux"
"Windows 2001"
"Windows 2001 SP1"
"Windows 2001 SP2"
"Windows 2006"
"Windows 2006 SP1"

I hope this situation isn't another foxconn scandal.

[wafflemelon]

ACPI can be fully loaded except for the thermal functions. I've had some success with kernel option:

thermal.off=1

Almikul, what are your current kernel opts?

[almikul]

ACPI can be fully loaded except for the thermal functions. I've had some success with kernel option:

thermal.off=1

Almikul, what are your current kernel opts?

My current options are

quiet splash acpi_osi=Linux acpi.power_nocheck=1

Do you know what exactly changed when you loaded thermal.off=1?

[derekmbarnes]

I hope this situation isn't another foxconn scandal.

Interesting find...I'm especially curious about this since my BIOS comes from AMI. While I don't want to falsely accuse anyone of wrongdoing, it would not surprise me at all to discover this apparent rigging process is still taking place. (For those who haven't clicked the link yet, do so.)

Which brings me to this question: could the problem ultimately be solved by installing Coreboot? I understand it's still a work in progress, but having a BIOS specifically built to support Linux systems would be a major boost to its freedom and functionality, not to mention possibly eliminating problems like those with Foxconn.

So armed with this knowledge, I'm going to reinstall Ubuntu on a virtual machine and see if I can isolate any specific problems with the DSDT. If I find it's biased against anything that isn't Windows, I'll call it out here, and then I'll start talking to the Coreboot developers to see if they're in a position to assist us. (This may take me several weeks as I'm shorter on free time.) Anyone else who's more prepared than I am is welcome to do the same.

Hope to return soon.

[wafflemelon]

I just had a chat with one of the guys at #coreboot @ freenode. Here's an excerpt:

[00:00] <CareBear\> Waffle : the shortest path to delivery is to get something already supported, or something identical to something already supported, or something as close as possible to something already supported
[00:00] <CareBear\> at the moment that means a ruggedised mil-grade laptop from Roda
[00:00] <Waffle> from what I've seen, the only laptops you have are olpcs
[00:00] <Waffle> not so much alike : )
[00:00] <CareBear\> I have a Samsung netbook which is getting old now, which is also likely to become successful at some point
[00:01] <CareBear\> Roda is a laptop as well, but not consumer-grade

So it seems it's unlikely that Coreboot will be a solution for us. It's a very interesting project, too bad it's difficult to include more mainboards.

I believe we only have the following options:

* Get Toshiba/Phoenix/whomever to create a GOOD version of their BIOS. It does look fishy to me. The Foxconn guy did it, so can we. Also, look him up, he has a blog

* Fix the DSDT file ourselves, recompile the kernel (custom DSDT is no longer supported) and create a patch for M500 owners. It takes some getting used to, but with some clever reading you can figure out how it works. Don't let the 14000 lines of code scare you ; )
By fixing, I mean either making the fan work exactly like it does for windows, or lowering all trip points by 10 C while making the fan work.

* Get the Linux/Ubuntu devs to do something about it. Don't ask me what. I believe this would be the longest and most unlikely alternative.

* Load the kernel with different options. I've ran out of ACPI options to feed to the kernel. I believe we already spent this one. If you feel like hitting this old nail, here's a list of options: http://redsymbol.net/linux_boot_parameters/
Reading the DSDT may help understand what you have to do.
Also, I find it very strange that acpi_os_name="Windows *" seems to have no effect for me.

It's a lot of work! If someone feels like going for either option 1 or 3, do tell us! Cramming some inboxes goes a long way : )

Derek, please share your views on the DSDT with us when you can!